Quick Buzz Feed
While the development of artificial intelligence dates back to the 1950s-60s, advances in the past decade have driven more recent developments in so-called generative AI, capable of producing text, code and audiovisual material based on patterns learned from large datasets. The deployment of AI can be divided into four key phases: innovation, development of new infrastructure, diffusion and widespread adoption of the new technology, and adaptation of business models and markets to the new technology.
Artificial Intelligence (AI) broadly refers to the capacity of machines and computational systems to mimic human intelligence for perceiving, synthesizing, and inferring information, enabling them to perform complex tasks previously requiring human cognitive abilities like language comprehension and decision-making. Though AI's roots trace back to the 1950s-60s, the last decade's breakthroughs in large language models (LLMs), coupled with enhanced processing power and extensive data collection, have propelled the rise of generative AI. This advanced form of AI can create text, code, and audiovisual content by learning from vast datasets. The implementation of AI unfolds across four key stages: initial innovation, the establishment of new supporting infrastructure, the widespread adoption and diffusion of the technology, and finally, the adaptation of existing business models and markets to incorporate these new AI capabilities.
The AI ecosystem is built upon a complex, interconnected value chain. This chain begins with critical minerals essential for manufacturing semiconductors, which serve as AI's 'physical brain'. These semiconductors are then integrated into a broader computational infrastructure that includes data centers, communication networks, power grids, and cloud computing services, forming the 'body' for large-scale data processing. Crucial to this infrastructure is access to vast datasets for training AI models. The subsequent link involves developing foundational models, demanding sophisticated algorithms and deep learning neural networks. Ultimately, generative AI's value is realized through creating specific applications, such as virtual assistants or content generation systems, built upon these foundational models, and their integration into digital products and services that directly engage end-users. The global landscape reveals significant variation in AI value chains across major economies. The US and some Asian nations, particularly Taiwan and South Korea, show high specialization in computational capacity, while the US has a more diversified chain. Conversely, China and several advanced economies, including Japan and the UK, lean towards specialization in computational capacity and data tools, with China and the Euro area emphasizing AI applications. This highlights a heterogeneous distribution of AI strengths across different phases of the value chain internationally.
The global economy is currently navigating the initial two phases of AI deployment: innovation and infrastructure development. Evidence points to a significant investment boom in these areas, particularly in the US and some Asian countries, driving exponential improvements in AI capabilities through 'hyperscaling'—rapid advancements in data volume, model parameters, and computational power. This surge in demand, however, is leading to infrastructure bottlenecks, especially in the chip, data center, and energy markets. Looking at the US, EU, and China, important distinctions emerge in the deployment process. The US holds a clear lead in the innovation phase, excelling in technological output, academic publications, open-source development, and chip design. China, however, is rapidly closing the gap, with its advanced models performing similarly to American counterparts, and showing strong innovation in patents and model development. The EU, by contrast, lags in innovation indicators, notably its low market share in chip production, making it dependent, and falling behind China and the US in AI model development. China leads in material supply due to its access to critical minerals and processing capabilities for chip manufacturing. In terms of AI adoption, diffusion, and adaptation, the picture is more uniform. Around 30% of the population in the EU and US are current AI users, while China is just under 20%. Despite a slight US advantage, all three economies show similar readiness for adoption, diffusion, and adaptation. China's manufacturing sector has undergone rapid modernization and 'robotization' through aggressive industrial policies and investment in infrastructure and human capital, positioning it to benefit significantly from AI diffusion as a global technology supplier. Meanwhile, the European and US economies' intensity in digital services places them as potential leaders in the adaptation phase, a crucial stage for determining AI's macroeconomic impact.
The 'AI race' is still in its nascent stages, with the US currently leading the innovation phase, but China is rapidly catching up, making it unlikely to be a two-horse race. Given AI's immense transformative potential, its successful deployment and macroeconomic impact will largely hinge on the business sector's capacity for adaptation and its ability to manage the associated frictions. Beyond the private sector, states must play an active role in regulating AI, promoting its adoption and diffusion, facilitating adaptation, and coordinating efforts globally. This involves fostering necessary improvements in institutions, infrastructure, and human capital. The challenges ahead are complex, demanding new public policy tools and economic diplomacy. Furthermore, the eventual global AI supply model—whether fragmented into blocs centered around the US or China, or more globally integrated—will have profound implications extending beyond just the economy. The AI marathon has only just begun, and all major players are actively participating.